Estimated Nonreimbursed Costs for Care Coordination for Children With Medical Complexity

We conducted a time study of care coordination activities performed for children and young adults with medical complexity enrolled with the University Hospitals Rainbow Center for Comprehensive Care from January 2013 through June 2015.

The University Hospitals Rainbow Center for Comprehensive Care (“the center”) is a multidisciplinary care coordination program established in 2013 and intended to address the Triple Aim¹⁷ for community-dwelling children and young adults with medical complexity in Northeast Ohio. Following the model of tertiary care–primary care partnership established by leaders in the field,¹⁰ the center teams with families, primary care clinicians (PCCs), and subspecialty services to optimize care and address barriers experienced by CMC and their families. Patients may be referred to the center (1) by an inpatient service at UH Rainbow Babies and Children’s Hospital, a free-standing children’s hospital aligned with a pediatric accountable care organization (ACO); (2) from the ambulatory setting by their PCC or specialist; or (3) by self-referral.

Children and young adults aged 0 to 26 years eligible for enrollment with the center experience significant neurocognitive impairment, involvement of 3 or more body systems, dependence on technology, and/or reliance on caregivers for instrumental activities of daily living (Tables 1 and 2). CMC are ineligible for the program if they are already served by a disease-specific care coordination program (eg, sickle cell, cystic fibrosis), if they are currently exercising their hospice benefits, or if they reside in a long-term care facility. There are no restrictions to enrollment on the basis of insurer. A total of 208 children and young adults with medical complexity were included in this study. At enrollment, these CMC were between 3 months and 26 years of age, with mean age of 8 years. Two-thirds demonstrated neurocognitive impairment. Three quarters relied on technology, including feeding tubes (69%); tracheostomies (15%); respiratory support devices including ventilators, continuous positive airway pressure, or bilevel positive airway pressure (9%); and other technology including central lines, urinary catheters, intrathecal baclofen pumps, vagal nerve stimulators, etc (24%).

Center staff function as a care coordination team linking CMC and their PCC with coordinated specialty services. Staffing assumes enrollment of 75 to 100 patients per full-time equivalent (FTE) clinician. The team consists of 0.5 FTE general pediatricians (MDs), 2 FTE pediatric NPs, 3 FTE registered nurse (RN) case managers, 2 FTE dietitians, 2 FTE SWs, and 4 FTE medical assistants hired as family care advocates (FCAs), a novel role intended to facilitate home visits with center enrollees. Center MDs and NPs are available 24 hours per day, 7 days per week. On enrollment, the center team in partnership with the CMC and their family develops a comprehensive care plan considering findings from detailed medical, nutrition, and psychosocial assessments. This care plan is made available to the center team, PCCs, and specialists via electronic medical record, and is used to guide care. Center clinicians see patients for both planned and urgent visits in the clinic and are available in some situations to conduct home visits. A center NP is available for consultation to inpatient teams to enroll new patients, to participate in multidisciplinary team rounds for established patients, and to assist with hospital discharge planning. Discharge from the program may occur if families move out of the ACO region or if they request program discharge.

This analysis included all recorded observations for all 208 children and young adults who enrolled with the center from January 1, 2013 (program start), through June 1, 2015. Data collection ended October 1, 2015.

Center staff documented patient characteristics at the time of program enrollment into a study-specific SharePoint Enterprise (Microsoft Corporation, Redmond, WA) database housed on Health Insurance Portability and Accountability Act–secure, hospital-based encrypted servers.

For each care coordination activity for each CMC, the individual performing the activity documented in the central SharePoint database their discipline (MD or NP, RN, dietitian, SW, or FCA); the method of the interaction (e-mail, telephone, chart review, video conference, or formal team review); the target of the activity (patient or caregiver; outside provider not employed at the center such as durable medical equipment suppliers, home nursing agencies, county Department of Children and Family Services, specialists, therapists, PCCs, etc; or other); and the duration (in 1 minute increments) of the activity. Direct data entry by each staff member into the SharePoint database occurred as close to real-time as possible, and no later than by the end of the work day in which the activity took place. To ensure consistency among staff, documentation of these activities was discussed as part of monthly interdisciplinary team meetings; however, the SharePoint database was not independently audited for accuracy. Activities omitted by this design included informal conversations among center staff outside the team case review setting and other communication activities such as faxing records. This study was approved by the University Hospitals Cleveland Medical Center Institutional Review Board.

The primary outcome was total time spent by center staff in care coordination activities. We estimated personnel cost by multiplying the hours in a given month spent in care coordination for a given child by the estimated hourly wage of the category of personnel performing the activity. We used estimated wages as reported by the Bureau of Labor Statistics Occupational Employment Statistics Survey, May 2015 (see Supplemental Table 5).¹⁸ We did not assess indirect costs or benefits associated with these wages. Because MD and NP times were recorded together by center staff as a single category, hourly salary for this category was weighted to reflect program time commitments by personnel type. Additionally, any in-person activities by MDs or NPs were excluded from these analyses because some might be in fact billable services depending on the setting in which they occurred.

Child-level factors of interest included sex (male or female), age, presence of neurocognitive impairment (yes or no), insurance status, and insurance type. Family-level factors included distance of residence from center location, family socioeconomic status at 4 levels (household income as percentage of federal poverty level: 0%–100%, 101%–250%, >250%, or missing), number of adult caregivers in the household, and receipt of food assistance (yes or no).

First, we described child- and family-level demographic characteristics for all children in the study. We next performed bivariate analyses using Pearson’s χ² or Fisher’s exact test (when expected cell size <20) to assess relationships between categorical child- and family-level characteristics, and the Wilcoxon-Mann-Whitney test to evaluate relationships between those characteristics and care coordination effort (time and cost per child per month). After estimating total time and direct personnel cost per child per month for all nonreimbursed activities, we constructed generalized linear models for each wage category to adjust cost estimates for child sex, age at enrollment, family income, and length of time enrolled in the program, censoring observations at date of program discharge, patient death, or end of the observation period. Sensitivity analyses tested for robustness of estimates to missing data and varying model assumptions. All analyses were conducted by using SAS 9.4 (SAS Institute, Inc, Cary, NC).

We recorded 53 148 care coordination activities for 205 CMC (3 enrollees contributed no activities before program discharge), an average of 19 327 activities per year and 21 activities per child per month. Our observations totaled 2586 child-months, ranging between 1 and 33 months for a given child (Table 3).

Although RNs and SWs accounted for a greater number of total activities, the clinical practitioners (MDs or NPs and dieticians) accounted for a larger proportion of time spent in such activities. Overall, all team members together spent on average 4.8 hours per enrolled child per month (median 2.3, interquartile range [IQR] 0.8–6.8) in care coordination activities (Fig 1).

Twenty percent of all recorded care coordination activities consisted of communication with a primary care MD, specialist, or ancillary provider outside the comprehensive care team. These communications required ∼10 minutes each (median 7, range 0–330). Across all disciplines, an average of 61 minutes per child per month were spent in communication with other providers (median 30, range 0–846). A slightly higher (21%) proportion of these communication activities involved either the patient directly or his or her caregiver. These communications were the most time consuming, requiring median 10 minutes for each communication (range 1–330), and totaling median 48 minutes per child per month (range 0–661). The majority of care coordination activities were classified as “other,” a category that included chart review and formal case review processes, weekly team-based review of new enrollees, and updates regarding acute concerns, emergency visits, and hospitalizations of established patients within the preceding week. Across all disciplines, care team members spent median 100 minutes per child per month in these activities (range 0–4060).

The majority of recorded care coordination activities comprised telephone calls, chart review, team-based case review, and (for non-MD and non-NP personnel) in-person communication with patients and their families. In contrast, little time was spent on e-mail (n = 5966, median 5 minutes per activity, range 1–120) or video calls (n = 122, median 5 minutes per call, range 1–90).

We found that program staff spent substantially greater time in care coordination for children in their first month of program enrollment than they did in subsequent months. In that initial month, program staff spent median 6.7 hours per child in care coordination activities (IQR 3.7–10.6). This compares with the more modest 2.2 hours per child per month (IQR 0.7–6.4) devoted in subsequent months (Fig 2).

Our second objective was to estimate personnel costs for these activities. We first estimated unadjusted personnel costs for care coordination. Assuming 25th percentile wages for all staff involved in care coordination, we found median cost per child per month to be $68 (IQR $19–$208), increasing to $98 (IQR $28–$300) if 75th percentile wages were assumed. Adjusting for family income, child sex, child age at program enrollment, and length of time with the program, we found that younger child age at enrollment was significantly associated with greater personnel costs for care coordination. In addition, compared with families with income exceeding 250% of the federal poverty level, those children whose families reported income at or below 100% of the federal poverty level were more likely to incur increased costs for care coordination. These observations were consistent across models assuming hourly personnel wages at the national 25th percentile, median, and 75th percentile, respectively. In sensitivity analyses, inclusion of presence of neurocognitive impairments or distance from home to the center did not improve model fit and were not significantly associated with personnel costs.

Therefore, accounting for child age, sex, duration with program, and family income, adjusted median personnel cost per child per month was between $145 (assuming 25th percentile wages) and $210 (assuming 75th percentile wages; see Table 4).

CMC are increasingly becoming the focus of attention nationally^4,15,19,20 both because we have much to learn about how best to manage their care and because it has become increasingly apparent that these children consume a substantial proportion of the fraction of Medicaid dollars that go to children. We found that in the context of a children’s hospital that is generally well staffed to care for these children, care coordination activities required conservative estimate of $145 to $210 in unreimbursed salary per child per month, not including fringe benefits, management or overhead costs. We note that our estimates are conservative, as our study design explicitly excluded important time-consuming tasks such as informal conversations among team members regarding a patient’s care and appointment scheduling. It is possible that over time the development of systems to support this work (for a cost) and efficiencies borne of experience or economies of scale could alter these salary costs. Nonetheless, salaries are likely to continue to constitute a significant cost to support providing coordinated care for CMC.

Care management requires work, and that work consumes resources. The value from care management may be generated through improved family- and patient-centered outcomes (including better health and greater satisfaction). Value may also be produced by increasing the efficiency of care; inappropriate services can be reduced, preventable complications avoided, and services may be provided in a more timely fashion. One challenge is that cost-efficient and even cost-savings programs may not benefit the provider of the care coordination services. For example, assuming 75th percentile wages for clinicians and staff, we estimate a median cost of just over $200 per CMC per month in direct salary costs to provide such services ($2400 per CMC per year); the equivalent cost of 4.2 emergency department visits (at median cost $570 per visit²¹), or 1 hospital day (at $2400 per day²²; see Supplemental Table 6). Thus, although the global outcome may be beneficial and efficient, there are market inefficiencies that allow for perverse consequences from critical perspectives; a clinical system may provide care coordination services whose efficiency reduces future revenue while a payer may benefit by avoiding future costs for no or minimal investment. Modern innovations, such as ACOs, and a population-based value framework¹⁹ attempt to create better alignment of incentives.

Whether for assessing potential costs and savings for health planning in an innovations framework or for budgeting in a more traditional framework, the field needs a better understanding of the costs and benefits of care coordination for this vulnerable and relatively expensive population. We provide a conservative estimate of the cost component.

The time study described in this report supports planning and design. High quality and safe care for these children requires care coordination.^23,–26 Our finding that new enrollees require substantially greater hours for care coordination in their first month compared with the observed subsequent relative “steady state” may help similar programs to pace admissions of new enrollees. In this era of accountability, developing payment models that align accountability, costs, and benefits will be critical for such care coordination efforts to be sustained by health care providers.²⁶ 

For its contributions, this study has important limitations. First, as an observational, uncontrolled assessment of clinical activity in a real-world setting, we have designed this study so that its biases reduce our estimate of costs. We considered only wages for time in direct services provision. We did not account for any of the additional costs required to maintain such a care program, such as program management, facility fees, and overhead. Our estimates of time and cost relied on staff logging of time spent in various activities. We speculate that when faced with pressing patient care concerns, staff may have forgone logging of those activities, further contributing to underestimation of total time expended in care coordination. Second, our program experienced substantial growth during the study period. It is therefore possible the program was over- or understaffed to meet the actual needs of CMC at various points during the study period.

We thus interpret our estimates as a low approximation of the nonreimbursed activities provided by our program. The data are specific to the design of our program; care coordination in other settings may require more or less staff. It is also likely that costs will not remain constant over time. Further research is needed to determine if our findings would hold in a more mature program. Finally, we have tried to make our estimates more general by using national salary estimates rather than local salaries. Although this study did not directly assess outcomes in terms of the Triple Aim, it does bridge the gap in the literature by estimating costs and efforts required to perform services that others have shown to be effective.^4,17,23 

This study aims to provide valid estimates of the unreimbursed direct salary cost of care coordination for CMC. At $145 to $210 per child per month, even this conservative estimate of costs represents a substantial investment by the institution that provides health care. Our data may be helpful for planners seeking to develop similar programs and those considering the costing of similar programs in the accountable care context, and for informing and stimulating conversations to develop evidence-informed fixes for the fee-for-service context.

ACO

accountable care organization

CMC

child(ren) with medical complexity

FCA

family care advocate

FTE

full-time equivalent

IQR

interquartile range

MD

medical doctor

NP

nurse practitioner

PCC

primary care clinician

RN

registered nurse

SW

social worker

We gratefully acknowledge Susan Connor, PhD, Nellie Coughlin, MSN, APRN, PPCNP-BC, Nicole Deutsch, MSN, APRN, PCNP-PC, Samantha Drake, RN, CPN, Katina Ferguson, LISW-S, Rebecca Loos, RN, Mary Neff, LSW, Laurel Rose, RN, CPN, Bernida Rowe, Amber Smith, LISW-S, Sarah Weiss, LISW-S, Jenna Wiltzer Karim, MS, RDN, Jodi Wolff, MS, RDN, and the patients and families of the University Hospitals Rainbow Center for Comprehensive Care for their contributions to this work.